Notes on AI, development, open source, and building in public.
An in-depth comparison of five computer-use projects: desktop control, visual agents, virtual machines, open-source licenses, security, cost and the best stack for personal AI.
A neutral, source-based comparison of OpenShip, Dokploy and Coolify: deployments, Docker Compose, clustering, CI/CD, databases, backups, MCP, security and costs.
A practical 2026 comparison of Patchright, Vercel's agent-browser and PinchTab: architecture, stealth, AI token usage, speed, reliability, security and when to use each.
Why I choose open source over proprietary software: transparency, user freedom, fair business models, less lock-in, and honest security trade-offs.
DeepSeek's October 17 update promises faster agents and shorter outputs. The benchmark gains, unchanged API prices, MIT weights and a practical API example.